Reproduce “Efficient Neural Surrogates for Linear Radiation Transport on the Lattice and Hohlraum benchmarks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Carmelo Gonzales, Steffen Schotthöfer, Cory D. Hauck
The authors train Transolver and BSMS-MGN surrogates to predict final-time particle-concentration fields for two two-dimensional linear radiation-transport benchmarks. They compare Fourier coordinate features and region-weighted loss across three seeds. Adding Fourier features lowers Hohlraum field RMSE for Transolver from 0.0559 to 0.0244, but raises it for BSMS-MGN from 0.107 to 0.115. These findings concern final-time fields on the trained mesh.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.